Data center infrastructure is a collection of devices known as the network — or the data path — that deliver applications and services to customers. In that data path lies a number of network and application services that provide for the scale, security, and speed of the applications they deliver.
In earlier days, data centers had been providing scalability mainly in two ways. The first way is data partitioning, which means storing data in different places. When a single database divides data across multiple partitions, each hosted on a separate server, you can scale out the system almost indefinitely. The second way to achieve a high level scalability is to create multiple identical app service instance copies. In this way, instance copies are organized into a cluster, where they work together so that, in many respects, they can be viewed as a single system.
With the reduced cost of hardware devices, wider usage of cloud computing, as well as the emergence of container technology, it is becoming increasingly harder for traditional data centers to meet the ever-changing requirements. On the one hand, some larger customers are complaining that app services delivered by a single data center cannot feed the huge number of requests from users; on the other hand, more and more enterprises in middle or small levels hope that app services can be charged by flow.
With changing requirements from customers, the way by which app services are delivered by the data center should also be changed accordingly. As a result, dynamic Infrastructure appears on the horizon. Especially in the mode of DevOps emerging in recent years, app services are beginning to be broken up into numbers of “Micro Services” functioning independent of each other. These micro services are orchestrated to app servers which can be delivered to final users. Some advantages of Micro services are obvious, with smaller size and less function, they are much they are more easily developed, tested and delivered; even more important is that, it provides scalability in a revolutionary way. Here app services are decomposed by function and each function is implemented by one or more micro services, which can be created or broken up dynamically. In this way, scalability can be made easier by dynamic infrastructure.
Let’s study two cases about how dynamic infrastructure makes scalability easier.
We assume that, an e-Commerce app service has been delivered to dynamic infrastructure. Initially, there is only one micro service dealing with order request from buyers. If more than10,000 requests are sent to this micro service simultaneously, there will be dramatic depreciation in performance. On a normal day, it can work well. However, if the day comes when many online stores offer huge discounts (such as “Double 11”in China), the number of order requests will skyrocket in a short time, maybe up to as much as 50,000! In such case, policies can be defined, that when there are excessive order requests coming, a new mirco service will be cloned and deployed.
Consider another case, an app service, used to manage finance data, is delivered by dynamic infrastructure. Under normal conditions, request for data will be forwarded to a data center in Beijing. However, if the data center in Beijing cannot unexpectedly work for some reason (known as a disaster), dynamic infrastructure will forward the requests to another data center located in Shanghai.
In both the above-mentioned cases, easier scalability has been achieved though dynamic infrastructure.
网友评论